Re: 2.4.7 softirq incorrectness.

[Maksim Krasnyanskiy]

> > Also it doesn't fix the scenario that I described (reschedule while running). I'm still wondering why don't I hit that trylock/BUG 
> > in tasklet_action.
>
>How old the problem is ? Was it always present ?
That's a good question. Data ordering check in Bluetooth tools was introduced pretty recently.
So, before 2.4.7 I wasn't paying attention to it and there for didn't notice any problems with 
tasklets.

>To be honest, this is too strong bug to believe to this at all. :-)
:) Agree. But I checked all my code, tx_task is called from tasklet only. And I do see that it's getting run on several 
cpus at the same time.

Also don't you agree with that it's possible (at least in theory) to hit that trylock/BUG in tasklet_action ?

Max

Maksim Krasnyanskiy	
Senior Kernel Engineer
Qualcomm Incorporated

[e-mail]
http://bluez.sf.net
http://vtun.sf.net